Scala 3 needs to settle for standards. We can’t keep changing the language. It’s going to kill it. Scala’s bigger problem is tooling, not the language itself. Stop reinventing the language and create more tools: a proper build tool, sbt is horrible
@HawaMahomed-z9k2 ай бұрын
Please cancel thanks.
@paulfunigga2 ай бұрын
That is black magic, literally. This talk is for people who are already experts in macros...
@nikolaychalkanov8963 ай бұрын
This guy gives me hope with scala at the right moment. I was about to give up. Everything around scala ecosystem is so complicated...
@einfacherkerl32793 ай бұрын
bootstrap and jquery?? they are already outdated
@vitaliiivanov95144 ай бұрын
That's really lots of cool stuff for DSL development. Convinced me to really learn Scala!
@JoanGoyeau4 ай бұрын
Really cool. Thanks Guillaume
@petacardi5 ай бұрын
Shame my brain is too small 😭
@mouadtribak16675 ай бұрын
Its always a pleasur to listen to this man
@PeatySpirit5 ай бұрын
I see, we are still on track to make Scala unappealing to industry as much as possible.
@andrewroby11304 ай бұрын
Sadly I think you're right. It's a pity that such a beautiful language doesn't get more love, but the comments I see most all come down to some version of broken or wonky tool chains. It's not the language proper, it's the ecosystem.
@blackflag.jobs_media5 ай бұрын
Really fantastic work!
@ganeshg89465 ай бұрын
Great talk !!! I love this person & his contribution to industry/society.
@pyepye-io4vu5 ай бұрын
Is it possible to use Storch to do GPGPU or parallel programming on the GPU in general? (Not deep learning) Like, say a GPU accelerated parallel blur filter or parallel merge sort for educational purposes? (I know this can be done on the CPU with parallel collections, etc., just want to show more speed-up with GPU.)
@scalevolvable3235 ай бұрын
Pretty talk! :)
@BillyLiuZhzh5 ай бұрын
Good Presentation, still missing Optimus. Demanding clients lol <- hopefully it is not me
@eg49336 ай бұрын
I strongly recommend learning scala for anyone interested in programming.
@mqtrade57435 ай бұрын
Why not go or java?
@Turing-mn1ys2 ай бұрын
@@mqtrade5743 The way Scala implements OOP is the latest and most developed way to implement it, if you understand programming language theory very well, you will see what Martin Odersky has delivered is truly a master piece.
@eg49336 ай бұрын
eventually scala will be the general go-to language replacing many other languages. It does both OOP and Functional in-house academically correct from getgo.
@programmer-z9b6 ай бұрын
You don’t use onComplete when dealing with Future in actors of Pekko/Akka,but you use pipeToSelf code block in which you define commands to be sent in both Success and Failure case, to the actor’s next behavior it will transition to.
@joan386 ай бұрын
16:19 Why is Spark/Databricks stuck on ancient versions of everything (Java, Scala...)?
@joan386 ай бұрын
19:40 I guess is the answer
@joan386 ай бұрын
Incredible!
@Filaxsan6 ай бұрын
16:00 Scala community, protect this guy at all cost. Great job (and principles) Mr. Li!
@esmaelmohamed82098 ай бұрын
Martin exceptionally Brilliant engineer yet he is very hamble!! we are as a community lucky having him, Respect!!
@mayonesacosmica9 ай бұрын
I was hoping to see how we could do this functionally but, alas, weights are mutated. 😢. But I see that the higher-level API could be enhanced so that `backward()` can return a loss function and `step()` can return a new optimizer. Is it not efficient enough?
@sorenbrunk98598 ай бұрын
Right now, the focus is on providing a very similar API to Python PyTorch, to ease porting models. But it would be great to add an immutable API (i.e. like JAX) as well. Contributions welcome!
@tojatos9 ай бұрын
Easy to digest knowledge - thanks!
@shumymikaball9 ай бұрын
Would it make sense to have the ownership/borrowing constraint model of Rust to Resources? Would this be possible to implement in scala?
@flwi9 ай бұрын
@rockthejvm That's a really great talk! Too bad I missed it (and you) in Madrid. Always enjoy talking to people who push the scala community forward. Keep on rocking the jvm!
@pookiepats9 ай бұрын
go run main.go
@tomascabrera19059 ай бұрын
Wow!
@tonyhahn19559 ай бұрын
I like Li Haoyi's work, but it's not easy to get started since the documentations seem to target experienced users. For example, I'd like to try scalatex and scala-site but can't even make the most trivial example work. And the github project seems abandoned, so can't get any support. Really a pity, since a minimal build.sbt and some commands how to build the 1st example could make it much more accessible.
@jackwislor10 ай бұрын
Will capture checking enable game development on par with C++ ?
@vicc888110 ай бұрын
Omg, this is great.
@abhijit-sarkar10 ай бұрын
"Gradle is such garbage" is not really a motivation for a new build tool. The first release of Gradle was in 2008, while Mill 0.1.0 was released in 2018. Most people in the Java ecosystem haven't even heard about Mill, let alone using it. When it gets used for 15 years, and in projects with millions of LOC, we can talk about how much, if at all, it stinks. There's just not enough data to make any sort of comparisons now.
@joan386 ай бұрын
You seem to base your opinion on how the adoption is going, just like a sheep follows the group. This arises the question: Are you even able to make your own opinion? No one seem to make sense out of Gradle or SBT. We are developers and we code Scala, Mill is plain Scala that I understand just like I understand any other Scala project. I don't know what command to run? I look at the code or even use IDE auto completion. It's simple and does not require a PhD to freaking build my project.
@abhijit-sarkar6 ай бұрын
@@joan38 You seem to be unable to comprehend complete sentences and get easily excited on seeing shiny things, just like a grade one child would do. The comment you replied to is my own opinion, of course. The point here isn’t how good or bad Gradle or SBT is, but how useful Mill is. Having used Mill, I found it slightly more useful than a POS. I do agree on one point though, building with Mill doesn’t require a PhD, or else you wouldn’t be here.
@joan386 ай бұрын
@@abhijit-sarkar "You seem to be unable to comprehend complete sentences" => Ok let's see: ""Gradle is such garbage" is not really a motivation for a new build tool" => Ok why? "The first release of Gradle was in 2008, while Mill 0.1.0 was released in 2018." => I understand it's new. In what is this measuring "how useful Mill is"? "Most people in the Java ecosystem haven't even heard about Mill, let alone using it." => I understand not many people use it. In what is this measuring "how useful Mill is"? "When it gets used for 15 years, and in projects with millions of LOC, we can talk about how much, if at all, it stinks." => I understand once the 2 previous points (maturity and popularity) are met then we can judge "how useful Mill is"? Are we offsetting the judgment onto others and following them? "There's just not enough data to make any sort of comparisons now." => So in the end you mentioned 0 concrete technical point of comparison in your comment. The judgment is all based on maturity and popularity. How is this a fair judgment of a piece of tech? This comment would work just fine on Gradle against Maven in 2008-2014. It's purely anti innovation and progression to me.
@abhijit-sarkar6 ай бұрын
@@joan38 Sorry, I don’t have the time or patience to read your essay, so, thank you for taking the time for whatever you wrote 😀
@MrDejvidkit10 ай бұрын
Thank you!
@MrDejvidkit10 ай бұрын
This talks are great. I love them, keep them going.
@MrDejvidkit10 ай бұрын
Thank you this is super helpful.
10 ай бұрын
This convinced me to try out mill for my next project, even though sbt kind of just works right now. I guess, as always, the biggest hurdle for new tools is the plugin ecosystem around it :)
@PlatinumDragonProductions99910 ай бұрын
Once I discovered Scala, I've been hooked ever since. It has replaced Python as my favorite language; even though I still love Python, Scala is the greatest for me. My only concern is that the library system is nowhere near as developer friendly. I spend a lot more time scrounging to find libraries to do specific tasks and often have to use Java (ick) libraries if I can't find a Scala one. One of Python's strengths is that I could literally type what I was looking for into my search bar and either find a working library, or the command that I needed to accomplish a given task. Even if I had never used the command or library before, I could quickly read the doc and be off and coding. This not only sped up my learning of the language and its capabilities, but also meant greater satisfaction in being able to complete my project without too much scrounging. I hope that Scaladex can develop into something to rival Pypi soon. I love Scala! 😀
@jacku368510 ай бұрын
"PromoSM"
@GadKateyOcansey10 ай бұрын
Nice lecture.
@hussain102110 ай бұрын
I hope one day scala migrates away from the jvm and become an independent language
@jonathanwinandy10 ай бұрын
True, however it's part of the value proposition, to be able to compose big system, with some of the parts that are already defined.
@distantforest24812 ай бұрын
Scala-Native exists and it currently has 4.5k stars on github.
@gerardbosch11 ай бұрын
Hi anyone, I'd like to ask whether type classes would help to solve the problem of adding new terms, when using ADT-encoding? Thanks!
@kevalan104211 ай бұрын
80k people learned Scala but only 870 views nowadays, is Scala in sleep mode?
@Bibhu-it1yiАй бұрын
Because Scala is Difficult . May be Billions People Go to Grammar School doesn't mean all are successful.
@MrDejvidkit11 ай бұрын
Is there link to the repot with the code?
@MrDejvidkit11 ай бұрын
Structural types = Magic. This will open a lot of possibilities.
@alaindevos402711 ай бұрын
Basics. How do i a "helloworld" in mill ? Using vscode. If this works i can continue.
@joan386 ай бұрын
object project extends mill.Module { def hello = T { println("Hello World") } } Then you run it with: mill project.hello
@MrDejvidkit11 ай бұрын
This really puts my thinking about polymorphic functions to another level. That etha expansion on polymorphic functions looks fantastic.
@BenHutchison11 ай бұрын
As you mention, intellij is the leading Scala idea and I use it daily. Still well ahead of Metals overall for me. However. The reality of Scala 3 dev right in IJ is an exercise in frustration. It's not more features that are needed. It's that the core features are more performant and reliable. Code completion often takes 10 secs to display, which is effectively unusable. Sbt imports take 1-2 minutes. I constantly pause to wait for the ide to catch up during dev.
@angeloceccato11 ай бұрын
Wow! Caprese is so very interesting project. Scala 3 is a very good next step. Another good thing to steal from rust is cargo and other tools, imho, i always go in throubles with sbt.
@MrDejvidkit11 ай бұрын
Yes Tuple elimination in this case is awesome.
11 ай бұрын
Useful, but to be honest it started to become hard to follow when refinement types started. Still very appreciated though!